LaxMotion: Rethinking Supervision Granularity for 3D Human Motion Generation

TL;DR

LaxMotion introduces a novel supervision paradigm for 3D human motion generation that emphasizes structural consistency over exact coordinate fitting, enabling better generalization and diverse motion synthesis without direct 3D pose supervision.

Contribution

The paper proposes a new framework that learns 3D human motion through relaxed supervision using global trajectories and 2D cues, improving generalization and diversity.

Findings

Achieves comparable or better performance than fully supervised methods.

Generates diverse, coherent, and semantically aligned 3D motions.

Reduces reliance on precise 3D annotations, enhancing scalability.

Abstract

Recent 3D human motion generation models demonstrate remarkable reconstruction accuracy yet struggle to generalize beyond training distributions. This limitation arises partly from the use of precise 3D supervision, which encourages models to fit fixed coordinate patterns instead of learning the essential 3D structure and motion semantic cues required for robust generalization. To overcome this limitation, we propose LaxMotion, a framework that synthesizes realistic 3D motions without direct 3D pose supervision. Instead of regressing toward exact coordinates, LaxMotion learns 3D motion as a consistent explanation of global trajectories and monocular 2D kinematic cues. We introduce a structured motion factorization together with a reformulated training paradigm under relaxed observability. This design is further supported by relaxed regularization objectives that enforce view consistent alignment, orientation coherence, and structural stability. Under this relaxed supervision paradigm, LaxMotion generates diverse, temporally coherent, and semantically aligned 3D motions, achieving performance comparable to or surpassing fully 3D supervised methods. These results indicate that shifting supervision from exact coordinate matching to structural consistency promotes stronger reasoning and improved generalization, offering a scalable and data efficient paradigm for 3D motion generation.